Biomedical Applications of Knowledge Discovery in Databases Research Pro iect Descriptions Project 1. Temporal knowledge discovery in diabetus mellitus and collagen diseases Abstract One of the important problems of knowledge discovery in medicine is that clinical databases include the large amounts of temporal information. This project aims at extracting temporal knowledge from clinical databases on diabetus mellitus and collagen diseases by using rule induction methods and hidden Markov models. Experimental results show that knowledge which medical experts implicitly use for medical expert reasoning are extracted from clinical databases. There are plans for the analysis to be extended to other analysis methods and databases. Organization: Tokyo Medical and Dental University, Medical Research Institute Contact: Shusaku Tsumoto Address: Medical Research Institute, Tokyo Medical and Dental University 1-5-45 Yushima, Bunkyo-ward Tokyo 113 Japan Email: tsamoto@computer.org URL: http://hal900.med.osaka-u.ac.jp Duration: 0.5 years Number of People: 3 Tools Developed: PROBTEMP Academic Disciplines: Medical Informatics Keywords: hidden Markov model, rule induction, clinical databases, temporal knowledge discovery Project 2: Bone-marrow post-transplant care application of data mining Organization: Fred Hutchinson Cancer Research Center, Clinical Research Division Primary Contact: Isabelle Bichindaritz Address: Fred Hutchinson Cancer Research Center Clinical Research Division, D5-360 1100 Fairview Avenue North Seattle, WA. 98109 Email: ibichind@fhcrc.org URL: http://www.aim.thcrc.org Duration: 2 years Number of People: 10 on project, 1 in data mining Tools: Developed: customization and combination of known algorithms Purchased: MSBN, C4.5, MLToolBox Academic Disciplines: artificial intelligence, case-based reasoning Abstract Our project is to build and evaluate a knowledge-based computerized decision-support system for bonemarrow post-transplant care on the WWW. Its main reasoning methodologies are case-based reasoning and rule-based reasoning. Data mining is used in this project in synergy with machine learning to automatically update the knowledge-base (learn new rules, refine existing rules) and to predict the evolution of a patient's state in an adaptive way. Keywords: decision-support, case-based reasoning, predictive data mining, rules induction Project Related Publications: Bichindaritz I, Kansu E, Sullivan KM.,"Case-based Reasoning in CARE-PARTNER: Gathering Evidence for Evidence-Based Medical Practice", European Workshop on Case-BasedReasoning, 1998 (in press). Bichindaritz I, Siadak M, Jocom J, Moinpour C, Kansu E, Donaldson G, bush N, Chapko M, Bradshaw JM, Sullivan KM., "Care-Partner: A Computerized Knowledge Support System for Stem Cell Post-Transplant Long-Term FollowUp on the World Wide Web", American Medical lnformaticsAssociation AnnualMeetin& 1998 (in press).

/lp/association-for-computing-machinery/temporal-knowledge-discovery-in-diabetus-mellitus-and-collagen-0h0bSpvZ02